This is the Third Notebook that we are presenting, In this notebook we are looking to do all of the suggestions given to us from the presentation and some of the next steps outlined in the presentation.
There are 3 big things that we want to do in this notebook. The first is we want to do is adjust the revenue and budget for the time value of money. After that we want to do feature selection using two different methods: recursive feature selection (brute force method using a library called boruta) and feature selection using ANOVA. We will be running this on two different data sets namely the NA one where we drop all NAs and the imputed data set. We are doing it on these data sets specifically because they were what we analyzed in the second notebook and saw interesting results there. There was a bit about ANOVA mentioned in the second notebook and all of that applies here as well. The reason that we are doing two feature selections is because we want to test the difference between them. We want to see what features they recommend in common and the differences. After which we will see which feature selection method performs better using error metrics. We would also ideally like to do multiple train/test splits as we know that the error rate heavily depends on the train/test split and give a huge range of 100% (MAPE) in some cases. We also recognize that this may not be possible computationally as it already take a significant amount of time to run the model once let alone running it multiple times.
We also did an ANOVA feature selection in notebook two but the reason we are doing it here is to compare feature selection models as Joe was interested in which one would perform better. Additionally the ANOVA feature selection in notebook two was on the dataset where the time value of money was not adjusted for, so we can see the impact that changing time value of money alone has on the results.
The first thing we must do as always is load all of the libraries that we will be using
library(readr)package 㤼㸱readr㤼㸲 was built under R version 4.0.3library(stringr)package 㤼㸱stringr㤼㸲 was built under R version 4.0.3library(tidyverse)package 㤼㸱tidyverse㤼㸲 was built under R version 4.0.3Registered S3 methods overwritten by 'dbplyr':
method from
print.tbl_lazy
print.tbl_sql
-- Attaching packages ------------------------------------------------------------------- tidyverse 1.3.0 --
v ggplot2 3.3.2 v purrr 0.3.4
v tibble 3.0.4 v dplyr 1.0.2
v tidyr 1.1.2 v forcats 0.5.0
package 㤼㸱ggplot2㤼㸲 was built under R version 4.0.3package 㤼㸱tibble㤼㸲 was built under R version 4.0.3package 㤼㸱tidyr㤼㸲 was built under R version 4.0.3package 㤼㸱purrr㤼㸲 was built under R version 4.0.3package 㤼㸱dplyr㤼㸲 was built under R version 4.0.3package 㤼㸱forcats㤼㸲 was built under R version 4.0.3-- Conflicts ---------------------------------------------------------------------- tidyverse_conflicts() --
x dplyr::filter() masks stats::filter()
x dplyr::lag() masks stats::lag()library(dplyr)
library(mice)package 㤼㸱mice㤼㸲 was built under R version 4.0.3
Attaching package: 㤼㸱mice㤼㸲
The following objects are masked from 㤼㸱package:base㤼㸲:
cbind, rbindlibrary(VIM)package 㤼㸱VIM㤼㸲 was built under R version 4.0.3Loading required package: colorspace
package 㤼㸱colorspace㤼㸲 was built under R version 4.0.3Loading required package: grid
Registered S3 method overwritten by 'data.table':
method from
print.data.table
VIM is ready to use.
Suggestions and bug-reports can be submitted at: https://github.com/statistikat/VIM/issues
Attaching package: 㤼㸱VIM㤼㸲
The following object is masked from 㤼㸱package:datasets㤼㸲:
sleeplibrary(plyr)package 㤼㸱plyr㤼㸲 was built under R version 4.0.3----------------------------------------------------------------------------------------------------------
You have loaded plyr after dplyr - this is likely to cause problems.
If you need functions from both plyr and dplyr, please load plyr first, then dplyr:
library(plyr); library(dplyr)
----------------------------------------------------------------------------------------------------------
Attaching package: 㤼㸱plyr㤼㸲
The following objects are masked from 㤼㸱package:dplyr㤼㸲:
arrange, count, desc, failwith, id, mutate, rename, summarise, summarize
The following object is masked from 㤼㸱package:purrr㤼㸲:
compactlibrary(tidyr)
library(ggplot2)
library(sf)package 㤼㸱sf㤼㸲 was built under R version 4.0.3Linking to GEOS 3.8.0, GDAL 3.0.4, PROJ 6.3.1library(sjmisc)package 㤼㸱sjmisc㤼㸲 was built under R version 4.0.3Learn more about sjmisc with 'browseVignettes("sjmisc")'.
Attaching package: 㤼㸱sjmisc㤼㸲
The following object is masked from 㤼㸱package:purrr㤼㸲:
is_empty
The following object is masked from 㤼㸱package:tidyr㤼㸲:
replace_na
The following object is masked from 㤼㸱package:tibble㤼㸲:
add_caselibrary(highcharter)package 㤼㸱highcharter㤼㸲 was built under R version 4.0.3Registered S3 method overwritten by 'htmlwidgets':
method from
print.htmlwidget tools:rstudio
Registered S3 method overwritten by 'quantmod':
method from
as.zoo.data.frame zoo library(openair)package 㤼㸱openair㤼㸲 was built under R version 4.0.3library(zoo)package 㤼㸱zoo㤼㸲 was built under R version 4.0.3
Attaching package: 㤼㸱zoo㤼㸲
The following objects are masked from 㤼㸱package:base㤼㸲:
as.Date, as.Date.numericlibrary(countrycode)package 㤼㸱countrycode㤼㸲 was built under R version 4.0.3library(ggmap)package 㤼㸱ggmap㤼㸲 was built under R version 4.0.3Google's Terms of Service: https://cloud.google.com/maps-platform/terms/.
Please cite ggmap if you use it! See citation("ggmap") for details.library(blscrapeR) ##needed to get index for adjusting inflation
library(Boruta)package 㤼㸱Boruta㤼㸲 was built under R version 4.0.3library(randomForest)package 㤼㸱randomForest㤼㸲 was built under R version 4.0.3randomForest 4.6-14
Type rfNews() to see new features/changes/bug fixes.
Attaching package: 㤼㸱randomForest㤼㸲
The following object is masked from 㤼㸱package:dplyr㤼㸲:
combine
The following object is masked from 㤼㸱package:ggplot2㤼㸲:
marginlibrary(mlbench)package 㤼㸱mlbench㤼㸲 was built under R version 4.0.3library(Metrics)package 㤼㸱Metrics㤼㸲 was built under R version 4.0.3This is the csv where all the NAs are included and we will just drop all NA values
Read in the data that we want to work with
data_na = read.csv("allMerge_clean_withNA.csv")
head(data_na)Do some basic class conversions
#converting classes
data_na <- mutate_if(data_na, is.factor, as.character())
data_na$budget <- as.numeric(data_na$budget)
data_na$Total.Revenue = as.numeric(data_na$Total.Revenue)
head(data_na)Extract the Year from the release_date column and store it in a variable called year
data_na$Year = str_extract(data_na$release_date, "\\d{4}")
head(data_na)Create a table that will give us the adjustment amount based on a base year of 2020
table = inflation_adjust(2020)trying URL 'https://download.bls.gov/pub/time.series/cu/cu.data.1.AllItems'
Content type 'text/plain' length unknown
downloaded 2.3 MBtableCreate a data frame with the values we need from the table
table <- as.data.frame(table)
table$adj_value2 <- ((100 + table$pct_increase)/100)
df <- table[,c("year","adj_value2")]
colnames(df) = c("Year", "adj_value") #changing name for left_join
dfMerge the data frame from above with this data frame based on the year
data_na = left_join(df, data_na, by = 'Year')
data_naConvert adjusted_revenue and adjusted budget to an integer, we do this because there are a lot of decimals in some cases as the adjustments are very specific
data_na$adjusted_revenue = as.integer(data_na$Total.Revenue/data_na$adj_value)NAs introduced by coercion to integer rangedata_na$adjusted_budget = as.integer(data_na$budget/data_na$adj_value)
data_na